Businesses are spending hundreds of billions of dollars on AI for the workplace. But getting employees to use the tools to their full potential is a huge task. How will companies make sure they see a return on investment? #AI #artificialintelligence #work #futureofwork #tech #technology #innovation #workplace #bot #unemployment #employment #companies ► Enjoying FT content? Get a daily slice of the very best FT journalism with FT Edit. Free for 30 days then just £4.99 a month See if you get the FT for free as a student (http://ft.com/schoolsarefree) or start a £1 trial: https://subs.ft.com/spa3_trial?segmentId=3d4ba81b-96bb-cef0-9ece-29efd6ef2132. ► Check out our Community tab for more stories: https://www.youtube.com/@FinancialTimes/community ► Listen to our podcasts: https://www.ft.com/podcasts ► Follow us on Instagram: https://www.instagram.com/financialtimes ► Follow us on Instagram: https://www.tiktok.com/@financialtimes
ADVERTISEMENT
:00 AI CEO IRONICALLY GULPING H2O during the interview 😮💨😬😅😅😅
Experimenting with AI workflows is a breeze using Pneumatic Workflow; the no-code approach is very effective.
The video hits the nail on the head about the "training gap." Companies are dumping millions into AI but no one knows how to use it. It's the same with individuals—we're all trying to figure out which tools actually add value. I almost burnt out trying to manage subs for Sora, Veo, and a dozen others just to stay current. Eventually just moved everything to omnely to access all those models + ElevenLabs and major LLMs in one place. It’s way cheaper than expensing 5 different pro plans to my boss lol.
"95% of Gen AI pilots fail" is such a brutal stat, but not surprising. The fragmentation is the real killer. You need one tool for video, one for voice, one for text... it's a logistic nightmare. I solved my own personal "pilot failure" by consolidating to omnely. Having Sora, Kling, Wan, and Nano Banana all in one dashboard makes it actually possible to build a workflow without drowning in subscriptions.
The FT analysis about abstract productivity claims vs. actual use cases is fascinating. Coca-Cola using it for just a Christmas ad? 😂. Real productivity comes from having a seamless stack. I stopped trying to force adoption of 10 different tools and just got omnely for the team. Accessing major LLMs, Sora, and ElevenLabs from one spot cut down the friction so much.
Very interesting. Like your correspondant pov. PS: the whole AI thing won't work, but it's a too scary thought to bring it to the table.
I love your videos!
This breakthrough is incredible, but watching it made me think about something I read in Selwyn Raithe's book . The book talks about how we celebrate each AI milestone without noticing we're actually witnessing the systematic replacement of human purpose. The scariest part isn't the technology itself, it's how willingly we applaud our own obsolescence.
This is a Really honest video. I’ve seen all of this first hand. Even in a tech company as a SWE getting people to adopt and use Gen AI is a struggle. You can’t stick an empty chat window in front of people and expect them to 10x their productivity.
This video looks like an ad for AI and Google
Workers know that the more they use AI for work, the faster they are kicked out the door! What is happening is a sort of passive resistance.
I agree and it is painful. I have to double and triple check the output from the AI tools that I am being pushed to use at work.
Talking to senior leaders is the most ridiculous idea - they are almost always full of 💩and if not they are clueless.
This is a massive bubble that will burst.
Really Good information. Thanks!
Here in India, it's a mandate for all to have at least 1 story on Jira board every sprint however useless they end up being 😂
Total seeing the quick ROI of AI at work. It is like, I don’t need a trainer in work as I can get all the answers to do all the hands on work. We basically do NOT NEED SEVERAL LAYERS OF MANAGERS OR DIRECTORS. Because the the AI tools replaces them.
Imagine any recent model (GPT, Claude, Gemini, Grok, DeepSeek) as a gigantic library, but it’s attended by a guy behind a door with a small window who asks you: ‘Hi! What can I do for you?’ What you extract from him depends entirely on you—how you ask and how you interact. The chat interface, while initially easy to use (because it’s in natural language), resembles the dreaded black screen with a blinking cursor waiting for you to code instructions to make your Commodore 64 do something as simple and seemingly useless as a red circle bouncing back and forth across the screen. Nowadays, nobody questions whether computers are useful or not. LLMs and AI have opened the door to a new era of computing and information processing. We’re just in the early days.
Every time I hear, "You aren't seeing the benefits because you're not using it right", especially from people who profit by teaching it, that's a red flag.
My take away is that companies have NO idea what AI can and should do and they are hoping you can figure it out for them